摘要: |
针对K近邻算法(KNN)在对偏向于某个样本点的未知点进行三角质心定位时定位精度变差的情况,提出了应用相关系数去匹配蓝牙信标iBeacon位置指纹库的室内定位算法。通过比较待定位点和位置指纹库中参考样点的相似程度,并进行数据差异显著性检验,来检验采集的待定位点数据与指纹库数据是否显著相关,然后取相关性较高的样本点进行加权平均匹配定位。实验结果显示,相关系数匹配位置指纹库算法可将2 m以内的定位精度从65%提高到92%,相较于传统的KNN匹配定位算法有着定位精度高、计算量小、定位时间短等优势。 |
关键词: 室内定位 蓝牙信标 位置指纹库 相关系数 显著性检验 |
DOI: |
|
基金项目:国家自然科学基金资助项目(61402063);重庆市科技人才培养计划(新产品研发团队)项目(CSJC2013KJRC-TDJS40012);重庆市高校优秀成果转化资助项目(KJZH14213) |
|
Indoor localization of bluetooth beacon position fingerprint based on correlation algorithm |
WANG Yanli,YANG Rumin,YU Chengbo,KONG Qingda |
() |
Abstract: |
K-Nearest Neighbor algorithm,or“KNN”for short,often simply uses triangle centroid algorithm to locate.However,once the backlog site is close to a certain sample point,the accuracy of location will be greatly affected.Therefore,this paper proposes a correlation coefficient matching iBeacon position fingerprint algorithm.Firstly,through comparing the similar degree between undetermined point and a certain sample point of received signal strength indication(RSSI) fingerprint,it uses the test of data difference significance to test whether the backlog site data and fingerprint data is significantly correlated.Then,it takes the weighted average of high correlation sample points to get the result.Experimental results show that the positioning accuracy within 2 m can be increased from 65% to 92% with the correlation coefficient matching fingerprint algorithm.Compared with traditional KNN localization algorithm,the proposed correlation matching algorithm has higher positioning precision,smaller calculation amount,and shorter positioning time. |
Key words: indoor positioning bluetooth beacon position fingerprint correlation coefficient significance testing |